Roland Potthast Webhome

• Sept 29 - Oct 3, 2025, Melbourne: International Symposium on Data Assimilation (ISDA 2025) https://isda-online.univie.ac.at/online-events/
• Sept 19 - Oct 11, 2025, Melbourne: visit

• Sept 2025, Switzerland: COSMO General Meeting COSMO General Meetings

• August 29-31, 2025, Heusenstamm: 50 years Freie evangelische Gemeinde Heusenstamm FeG Heusenstamm

• July 7-11, 2025: EUMETNET E-AI Workshop on AI/ML for Weather and Climate Products and Services https://www.eumetnet.eu

• June 14 - 29, 2025: reflection time

• May 26 - 28, 2025, Offenbach: ML NWP Python Days II
• May 14, 2025, Offenbach: Visit of Dr. Takemasa Miyoshi at DWD Profile

• April 14-16, 2025, Offenbach: ML NWP Python Days I
• April 9 - 11, 2025, Bonn: 50 Years ECMWF Seminars Website

• March 31 - April 4, 2025, Offenbach: ECMWF Machine Learning Pilot Project Meeting Previous Event

• March 26-29, 2025, Norway: Workshop on Machine Learning for Weather Prediction
• March 17-21, 2025, Bonn: PrePEP Conference “Precipitation Processes - Estimation and Prediction (PrePEP)” Website
• March 13, 2025, ISDA-Online: Data Assimilation Methodology Session ISDA Online March 2025
• March 10-11, 2025, Offenbach: ICCARUS Workshop Website

• Feb 21 - Mar 5, 2025, Sapporo, Japan: Scientific Visit to Prof Gen Nakamura Gen Nakamura
• Feb 18-19, 2025, Berlin: national ESM Workshop (natESM) natESM Workshop 2025
• Feb 13, 2025, ISDA-Online: Session on Coupled Data Assimiation ISDA Online Feb 2025
• Feb 5, 2025: ICON Board Meeting https://www.icon-model.org

• Jan 29-30, 2025, Vienna: EUMETNET Programme Manager Meeting, WFC and E-AI Programme Management https://www.eumetnet.eu
• Jan 9, 2025, ISDA-Online: Special Session on Prediction Science ISDA-Online Jan 2025

Previous News News RP 2024, News RP 2023, News RP 2022, News RP 2021, News RP 2020, News RP 2019, News RP 2018, News RP 2017, News RP 2016.

Our department FE1 for Numerical Weather Prediction (NWP) of the German Weather Service (DWD) consists of about 110-120 scientists at DWD headquarters in Offenbach close to Frankfurt/Main and in Potsdam close to Berlin, about 60 of them on funded R&D projects. The department consists of four sections, with 20-35 scientists each:

• Data Assimilation and Predictability (FE11),
• Observations Modeling and Verification (FE12),
• Numerical Modeling (FE13),
• Physical Processes and Dispersion (FE14).

DWD is a part of the German Ministery of Transport and Digital Infrastructure (BMVI), but our work is carried out in intensive cooperation with

• the Max-Planck-Institute for Meteorology MPI-M, see also ICON,
• the German Climate Computing Center DKRZ,
• the Karlsruhe Institute of Technology KIT, in particular with ICON-ART,
• ETH Zurich, in particular C2SM,
• the weather services of the COSMO Consortium (Italy, Switzerland, Russia, Poland, Romania, Greece and Israel),
• the Meteorological Service of the German Armee Geo BW, see also PDF,
• the Hans-Ertel Center HErZ research branches at the Universities of Munich, Bonn, Frankfurt, Hamburg, Berlin

and further international partners such as the

• European Centre for Medium Range Weather Forecasting ECMWF located in Reading, Bologna and Bonn.
• EUMETSAT, in particular the NWP-SAF Consortium with the Partners MetOffice (UK), Meteo-France, ECMWF and DWD.

We are working on modeling and data assimilation for numerical weather prediction (NWP) and earth system simulation (ESM). Our main task is to provide operational data assimilation and forecasting with

• ICON-global, i.e. a global NWP-model with 13km resolution, 120 layers 75km height, run every 3 hours 24/7,
• ICON-EU, i.e. its mesoscale two-way-nesting area over Europe with 6.5km resolution and the
• ICON-D2, i.e. high-resolution convection-permitting analysis and forecasts over central Europe with 2km resolution, 24km height analysis every hour 24/7, forecasts every three hours.

We prepare a rapid update cycle (RUC) with forecasts every hour in integration with Nowcasting techniques. This system is called

• ICON-RUC and SINFONY, see also SINFONY Webpage.

Our development and services include ensemble data assimilation for the ensemble prediction systems

• ICON-EPS global with 26 km resolution and 40 members and
• ICON-EU-EPS with 13km resolution over Europe with 40 members as well as
• ICON-D2-EPS with 2km resolution over central Europe, 40 members.
• ICON-D1 and ICON-D05 with 1km resolution and 500m resolution is under development.

We run a

• hybrid ensemble-variational data assimilation scheme (EnVar) globally, i.e. a variational data assimilation scheme, coupled with an Localized Ensemble Transform Kalman Filter (LETKF).
• For ICON-D2 we employ the four-dimensional version of the Localized Ensemble Transform Kalman Filter (4D-LETKF).

Further,

• Particle filters, in particular the Localized Adaptive Particle Filter LAPF and Localized Mixture Coefficients Particle Filter LMCPF are available for global and regional scale and are being used for research within the operational framework.
• EnVar for convective-scale ICON-D2 is under development,
• 4D-EnVAR for both global and regional systems
• an ultra-rapid data assimilation scheme (URDA) based on our ensemble and also
• a coarse EnVar (cEnVar) using our global ensemble for regional ensemble data assimilation world-wide are being developed and tested.

Data Assimilation includes the use of a broad varity of both direct and remote sensing measurements from

• Ground Stations and Ships (SYNOP),
• Radio Sondes (TEMP) and dropsondes,
• Buoys,
• Air Planes (AMDAR, AIREP, ACAR, …),
• Atmospheric Motion Vectors (AMV),
• Scatterometers (SCAT),
• Infrared Sounders (IR),
• Microwave Sounders (MW) and Microwave Radiometers (MWR),
• LIDAR including Clound Bottom Height (CBH), Cloud Top Height (CTH), Backscatter Profiles, Line-of-Sight Winds (AEOLUS), Ceilometers,
• RADAR including RADAR Radial Winds, RADAR Reflectivity and RADAR Dual Polarization,
• GPS/GNSS including Radio Occultations (RO), Zenith Total Delay (ZTD), Slant Total Delay (STD),
• Cameras,
• Cars.

Geostationary satellites and polar orbiting satellites are used operationally, while a lot of research is going into the better use of hyperspectral observations (many thousand frequencies per observed atmospheric column) in particular over land and in cloudy situations. The observation and reconstruction of snow, ice, sea surface temperature, land surface temperature, coverage, emissivity and soil moisture is a very active area of research. Also, the observation and data assimilation of clouds and convective processes with high-impact phenomena such as thunderstorms, heavy rain and wind gusts with lead times from minutes to days is a special focus of our research.

The research of our group at the University of Reading, UK, is concerned with inverse problems and data assimilation in three areas:

• numerical weather prediction (NWP),
• cognitive neuroscience / neural field theory (NFT),
• inverse scattering problems / remote sensing.

These are extremely exciting areas scientifically and very important for society, for example for air traffic control, severe weather warnings and national energy supply, in medicine by medical imaging and for many industrial and environmental questions.

Since October 2020 I am heading the department on Numerical Weather Prediction (NWP) of DWD with about 110-120 researchers in-house (state June 2024). There are four division heads in this department to lead sections of approximately 20-35 researchers each, see group.

Thinking about faith and life has always been a passion for me. I have become a Christian and have started to explore the world as someone who follows Jesus - that has turned out to be quite an adventure and highly exciting. In my daily blog I explore thoughts and arguments about faith, and monitor how faith works on a daily basis: Jesus Network

Recent publications can be found on publications. A book Inverse Modeling by Nakamura and Potthast with an introduction into data assimilation and inverse problems has recently appeared at IOP.

Working in an operational center, our focus is to develop state-of-the-art high-dimensional analysis and simulation methods which can be run in a reliable way on a supercomputer in near real-time. It includes codes simulating atmospheric processes, fluid-dynamics, physical parametrizations, scattering of waves, propagation of light and radiation, tomography, large-scale optimization and uncertainty quanitfication, ensemble and particle methods.

With a SX-Aurora TSUBASA C401-8 of NEC our supercomputers at Deutscher Wetterdienst are no 101 and no 136 (June 2024) List on the TOP-500 Supercomputer List. We have access to Cray XC40, Xeon E5-2695v4 of ECMWF, Mistral - bullx DLC 720, Xeon E5-2680v3 12C of DKRZ and Piz Daint - Cray XC50, Xeon E5-2690v3 12C 2.6GHz, Aries interconnect, NVIDIA Tesla P100 of CSCS in Switzerland.

High-performance computing is our daily business, however, the development of insight into the scientific problems we need to solve is an indispensable ingredient of our daily work. Part of this insight is based on mathematical analysis and the testing of computational methods for purpuse-built small-scale demonstration systems.

• Frontiers in Applied Mathematics and Statistics - Dynamical Systems
• Mathematical Inverse Problems

• Steering Committee (STC) COSMO Consortium COSMO Structure
• Steering Committee ICON Consortium https://icon-model.org
• Steering Committee German national Earth System Modeling natESM Steering

• Programme Manager EUMETNET Programme Artificial Intelligence for Weather, Climate and Environmental Applications WMO Announcement

• Steering Committee High-Performance Computing Center Stuttgart https://www.hlrs.de/, HLRS
• National High-performance Computing (NHR) KIT representative Earth System Science https://www.scc.kit.edu/dienste/nhr.php
• National High-performance Computing (NHR) User Committee NHR User Committee
• Gauss Centre for Supercomputing Steering Committee https://www.gauss-centre.eu/, Gauss Steering Committee